REPLY:

This contribution is a reply to the comments of Britton et al. on our recent article in The Journal of Nuclear Medicine describing the potential use of ^99mTc-labeled human defensins 1–3 for discrimination between infections and sterile inflammatory processes in laboratory animals (1). The major criticism deals with our remarks concerning ^99mTc-labeled ciprofloxacin, which is mentioned in a single sentence in the introductory section of our publication. When submitting our manuscript, we were already involved in experiments comparing the possibilities and limitations of antimicrobial agents, including antimicrobial peptides (1,2) and ciprofloxacin (3), for the specific detection of experimental infections in mice. The results available to us at that time contributed to our creating this now disputed sentence. The publication by Vinjamuri et al. (3) was intended to acknowledge the work with ^99mTc-labeled ciprofloxacin, and the reference to the publication by Bäumler and Heffron (4) in this sentence should have been placed immediately after the mentioning of antibiotic-resistant microorganisms. As Britton et al. indicated in their letter, the reference to Vinjamuri et al. (3) was not optimal in the context of our remarks, and we apologize for this shortcoming. We do want to stress that it was never our intention to undermine ^99mTc-labeled ciprofloxacin for imaging but to make some critical remarks on the use of this pharmaceutical for discrimination between infections and sterile inflammatory processes.

We disagree with Britton et al.'s statement that we misled The Journal of Nuclear Medicine and its readers with our critical remarks. In agreement with data reported by others (5), we have found that this radiolabeled compound binds equally well to purified DNA from gram-positive and gram-negative bacteria (12 ± 2 μg of labeled ciprofloxacin/mg of DNA) and human leukocytes activated in vitro by bacterial products (13 ± 5 μg of labeled ciprofloxacin/mg of DNA), with the respective values for radiolabeled human defensins 1–3 being 0.07 ± 0.02 μg of labeled peptide/mg of DNA and 0.05 ± 0.02 μg of labeled peptide/mg of DNA (n = 3). In addition, in vitro binding studies revealed more binding of ^99mTc-labeled ciprofloxacin to activated human leukocytes and cytokine-activated human endothelial cells compared with Staphylococcus aureus, Klebsiella pneumoniae, and Candida albicans (Table 1). In contrast, radiolabeled human defensins 1–3 bind to human cells less well than to these microorganisms (Table 1). These data indicate that binding of radiolabeled defensins to bacteria and C. albicans involves other binding sites than DNA. Indeed, it has been reported that defensins bind to negatively charged residues in the plasma membrane and possibly to specific receptors as well (2). Because radiolabeled human defensins 1–3 preferentially bind to microorganisms, the suggestion by Britton et al. that, because of their size (approximately 4 kD) defensins will be retained more likely in a nonspecific fashion at the site of infections than ciprofloxacin, is not solid.

Furthermore, studies in mice have demonstrated similar values for the accumulation of radiolabeled ciprofloxacin in foci of infection and in sterile inflammatory lesions in mice, again in contrast to our results with human defensins 1–3 (Table 2). On the basis of the values found for ciprofloxacin in mice with an inflammation induced by lipopolysaccharides (LPS) and in mice having an inflammatory process induced by heat-killed bacteria, we concluded that radiolabeled ciprofloxacin most likely binds to infiltrating leukocytes and bacteria in the inflamed/infected area. Accumulation of ciprofloxacin in sterile inflammatory lesions has also been reported by others (6). Other disadvantages of (radiolabeled) ciprofloxacin are its ability to stimulate cytokine production in human cells (7) and the emergence of antibiotic-resistant bacteria (8). In connection to the latter, the (frequent) use of suboptimal concentrations of antibiotics as used by the group of Britton et al. is certainly a risk factor because many bacteria survive and, on contact at the site of infection, have the opportunity to become resistant to the antibiotics. No data on the development of ciprofloxacin-resistant bacteria in the studies by Britton et al. have been reported yet.

The second line of criticism by Britton et al. focuses on limitations of ^99mTc-labeled human defensins 1–3 for infection detection. Defensins are small cationic antimicrobial peptides with broad-spectrum antimicrobial activity against a variety of bacteria, fungi, and enveloped viruses. The possibility that defensins (as well as various other antimicrobial peptides) can leave the circulation to reach the site of infection where they preferentially interact with bacteria (and fungi) makes these peptides our candidates of choice for the development of new radiopharmaceuticals for infection detection (1,2). Indeed, we demonstrated that imaging of infections can be achieved with ^99mTc-labeled defensins (1). However, we realize that because of their tertiary structure, it is difficult to prepare defensins in large amounts under good manufacturing practice conditions. For this reason and because immunologic side effects could not be excluded (9), we concluded that defensins are not optimal for infection detection (2). This conclusion led us to study the possibilities and limitations of other antimicrobial peptides for imaging of infections (2). The occurrence of bacteria that is resistant to antimicrobial peptides, defensins in particular, has been reported (10). On the other hand, these peptides have been successful in protecting plants, animals, and humans against the continuous threat by a variety of pathogens (11), which argues against the suggestion by Britton et al. that resistance to antimicrobial peptides may develop easily. We do not agree with these authors that information about the behavior of defensins 1–3 in experimental animals is lacking (9). Moreover, we have found substantial accumulation of radiolabeled defensins 1–3 in infectious foci caused by an injection with various gram-positive bacteria (e.g., multi–drug-resistant S. aureus, Listeria monocytogenes), gram-negative bacteria (K. pneumoniae, Escherichia coli, Salmonella typhimurium), and the fluconazole-resistant C. albicans (Table 2).

In contrast to the statements of Britton et al., but on the basis of our results (showing small standard errors time after time), we have no doubts about the reliability of our results with radiolabeled peptides. In several articles, we already described in detail our method of labeling peptides with ^99mTc, including quality controls and stability of the radiolabeled peptides (2). In addition, to exclude the possibility that the labeling methods affect the ability to identify infections, we copied precisely the protocol of Vinjamuri et al. (3) for labeling of ciprofloxacin with ^99mTc and analyzed its capacity to accumulate at sites of infection in mice. The results for ciprofloxacin radiolabeled by our method and the method of Vinjamuri et al. with respect to infection detection were identical. Furthermore, ^99mTc-labeled defensins 1–3 rapidly accumulated in the kidneys, bladder, and liver, and no activity was observed in the thyroid and stomach of the mice (1), indicating little or no free pertechnetate in our radiolabeled antimicrobial peptide preparations. Currently, we are investigating the half-life of antimicrobial peptides (1) and their degradation in bacterially infected mice.

We were surprised by the comment of Britton et al. that the peritoneal cavity model is not useful to study the binding of antimicrobial agents to bacteria and leukocytes. The publication cited by these authors does not deal with binding studies but describes the inducible expression of rat enteric defensin genes. Therefore, this comment is of no concern to our study. Moreover, the reliability of the peritoneal model for binding studies of radiolabeled compounds has been described before (12). Our approach using radiolabeled antimicrobial peptides for infection detection is reproducible, because the standard errors of the mean in our results were rather small (1,2) and identical results were found by an independent research group (2). Furthermore, the peptides specifically identified infections, as we found no target:nontarget (T–NT) ratio values >1.5 in mice having a sterile inflammations (i.e., that had been injected with LPS and heat-killed microorganisms) (Table 2) or serum and phorbol myristate acetate (data not shown). In addition, we found a positive correlation between the number of viable bacteria in infected thighs and the T–NT values for defensins (1), as we also observed earlier for IgG. These data imply that the efficacy of the antibiotic treatment of infections can be monitored with radiolabeled antimicrobial peptides. It should also be noted that we never observed cytotoxic effects of the amounts of radiolabeled antimicrobial peptides we used, including defensins 1–3, in our animal experiments. Nevertheless, we are still studying new peptides for infection detection and taking possible side effects into account. Of course, extensive toxicity studies have to be performed with selected peptides before they can be taken into clinical studies.

Together, the major limitations of 2 new radiopharmaceuticals—ciprofloxacin and defensins 1–3—for infection identification are discussed in this reply. Our final goal is to develop optimal radiopharmaceuticals that discriminate between infections and sterile inflammatory processes. Therefore, we thank Britton et al. for their comments, which will enable us to provide more information on the possibilities and limitations of radiolabeled ciprofloxacin and human defensins 1–3 for infection detection. We remain open for further discussion on this important issue.